Connections.AI

Inspiration

We built this model as it seemed like the most fun to develop a solution for as the game itself is very enjoyable.

What it does

It makes use of both word2vec and FastText in order to develop a high-confidence level solution to the Connections Game.

How we built it

It makes use of both word2vec and FastText, as research has proven that LLMs cannot solve the connections word problem due to its sheer complexity.

Challenges we ran into

We ran into problems with the confidence level of the AI itself and its ability to choose the correct words.

Accomplishments that we're proud of

We're proud how efficiently we were able to overcome the challenges at hand. We were also great at working together as a team to get segments of the problem done.

What we learned

We learned a lot about different datasets regarding word connections.

What's next for Connections.AI

The methodology to receive the uncertainty quantification and get a higher confidence level.

Built With

• gensim
• llm
• python
• word2vec